Errors associated with three methods of assessing respirator fit.

Three fit test methods (Bitrex, saccharin, and TSI PortaCount Plus with the N95-Companion) were evaluated for their ability to identify wearers of respirators that do not provide adequate protection during a simulated workplace test. Thirty models of NIOSH-certified N95 half-facepiece respirators (15 filtering-facepiece models and 15 elastomeric models) were tested by a panel of 25 subjects using each of the three fit testing methods. Fit testing results were compared to 5th percentiles of simulated workplace protection factors. Alpha errors (the chance of failing a fit test in error) for all 30 respirators were 71% for the Bitrex method, 68% for the saccharin method, and 40% for the Companion method. Beta errors (the chance of passing a fit test in error) for all 30 respirator models combined were 8% for the Bitrex method, 8% for the saccharin method, and 9% for the Companion method. The three fit test methods had different error rates when assessed with filtering facepieces and when assessed with elastomeric respirators. For example, beta errors for the three fit test methods assessed with the 15 filtering facepiece respirators were < or = 5% but ranged from 14% to 21% when assessed with the 15 elastomeric respirators. To predict what happens in a realistic fit testing program, the data were also used to estimate the alpha and beta errors for a simulated respiratory protection program in which a wearer is given up to three trials with one respirator model to pass a fit test before moving onto another model. A subject passing with any of the three methods was considered to have passed the fit test program. The alpha and beta errors for the fit testing in this simulated respiratory protection program were 29% and 19%, respectively. Thus, it is estimated, under the conditions of the simulation, that roughly one in three respirator wearers receiving the expected reduction in exposure (with a particular model) will fail to pass (with that particular model), and that roughly one in five wearers receiving less reduction in exposure than expected will pass the fit testing program in error.